Unverified50% confidenceFactExact time
In DeepSeek-V4.1-Flash, only four layers are responsible for writing the compressed KV, and the rest of the model shares this cache.
1
Sources
50%
Confidence
Long-term
Relevance
9/15/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedQuantized KV caches (such as storing KV vectors in FP8 or INT4) can compress the cache size to 1/2 or even 1/4 of the original.70% similarUnverifiedTowards AI ultimately chose to use DeepSeq V4 Flash with hybrid retrieval, with a memory strategy of retaining as much as possible and setting a 30K token compression threshold as a fallback.70% similarUnverifiedGQA allows multiple query heads to share the same set of KV heads, compressing the KV cache size to 1/4 to 1/8 of the original.67% similarUnverifiedChoosing Q4_0 for KV cache quantization can compress the cache size to approximately one-quarter of FP16, with far less impact on output quality than model weight quantization.66% similarVerifiedMLA compresses KV pairs into a low-dimensional latent space for caching and recovers the full Keys and Values via an up-projection matrix during inference, compressing the KV cache by several times.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/921484API
curl https://kongchang.com/api/v1/knowledge/claims/921484MCP
get_claim(id=921484)